The invention relates to a method for determining an actuating variable for actuating a torque control element in an electric vehicle steering system, wherein the actuating variable is determined as a function of a predefined target torsion-bar torque.
The invention further relates to a control device for controlling a steering system in a vehicle, wherein the steering system comprises a torque control element, and wherein the control device comprises means for actuating the torque control element by means of an actuating signal and means for determining the actuating signal as a function of a predefined target torsion-bar torque.
The invention further relates to a computer program that can be executed on a control device for controlling a steering system, and more particularly on a microprocessor in the control device.
With electric servo steering, an actuating force is applied by a driver to a steering means, such as a steering wheel, so as to change or maintain a current steering angle. Using an electric motor, an assistance force or an assistance torque is generated, which amplifies the actuating force applied by the driver. The assistance torque thus determined is stabilized using additional measures, for example so as to prevent undesirable sudden changes in torque.
The actuating force, or the actuating torque, can notably be amplified as a function of the actuating torque so that less amplification is implemented, for example, when an actuating torque is particularly high, than when the actuating torque is low, and conversely.
A method is known from DE 101 15 018 A1 for operating an electric power-assisted steering system, wherein an actuating signal for actuating the electric motor is determined from the current state of the steering system by means of a state regulator so as to implement torque-free steering. To this end, a state observation element is connected upstream of the state regulator and monitors the state variables of the steering system, and based thereon determines input variables that are suitable for the state regulator. The state variables of the steering system that are used include a pinion-side angle velocity, an angle difference between the steering wheel and pinion side, an angle velocity difference between the steering wheel and pinion side, and a current torque generated by the electric motor. The actuating variable thus determined allowed torque-free steering to be realized, but initially disturbance variables were not considered. So as to ultimately allow disturbance variables to be considered as well, a disturbance variable observation element for determining disturbance states is provided in the known steering system, wherein the torque applied by the driver, or friction moments acting on the steering system, are regarded as the disturbance states.
The final actuating signal is then determined based on the actuating signal that allows torque-free steering, by feeding forward the disturbance variables and feeding forward a predefinable target torsion-bar torque. So as to avoid complex measurement of the state variables in the steering system, or so as to allow states that cannot be measured to be taken into consideration, a model of the steering system is used for estimating the non-measurable state variables. The estimated state variables of the steering system thus correspond to the behavior of the model of the steering system.
The accuracy of the actual torsion-bar torque, which is adjusted by means of the known system, thus decisively depends on the measurement or estimation of the system state variables and the disturbance variables. The disturbance variables, or the torque that is determined as a function of the disturbance variable observation element and fed forward for determination of the actuating variable, compensates for deviations from the assumed model behavior, in the real steering system. The accuracy of the determination of the actuating variable is thus dependent on the quality of compensation for deviations from the model behavior by the real steering system. Moreover, the accuracy is dependent on the quality of the selected model.